惯性导航系统在开始工作时需要进行初始对准从而确定初始姿态。提出了一种与经典的对准算法如陀螺罗经或卡尔曼滤波技术不同的凝固惯性系快速(IF3)对准算法。可在任意初始误差条件下进行对准,且能适应高频扰动环境。将姿态矩阵分解成地球自转、惯性速率和对准矩阵三个部分。对准矩阵依靠两组分别处于不同惯性系里的观测向量确定。通过采用前置平滑滤波、层叠采样和二重积分技术,对准精度显著改善。在载车发动机怠速运行和人员上下车扰动条件下,60 s 对准误差优于 1 mil (1s ),180 s 对准误差优于 0.6 mil(1s ),300 s 对准误差优于 0.4 mil(1s )。实验结果证明了 IF3 对准算法的快速性、准确性和鲁棒性。 Inertial navigation system at the beginning of the work needs initial alignment to determine the initial attitude. A fast freezing (IF3) alignment algorithm is proposed which is different from the classical alignment algorithms such as gyrocompass or Kalman filter. It can be aligned under any initial error condition and can adapt to high-frequency disturbance environment. The attitude matrix is ​​decomposed into three parts: the Earth’s rotation, the inertial velocity and the alignment matrix. The alignment matrix depends on two sets of observation vectors in different inertial systems respectively. Alignment accuracy is significantly improved by using pre-smoothed filtering, stacked sampling and double integration techniques. The 60-s alignment error is better than 1 mil (1s), the alignment error of 180 s is better than 0.6 mil (1s) and the alignment error of 300 s is better than 0.4 mil under the conditions of vehicle engine idling and personnel getting on and off. 1s). The experimental results show the fastness, accuracy and robustness of the IF3 alignment algorithm.